Calculating Functionals of Solutions of Large, Sparse Systems

Problems in many applications lead to large, sparse linear systems with coefficient matrices that are invertible and have little other structure. In such problems, the solution u=Af is typically calculated only to compute further functionals of that solution. This paper performs preliminary research into the practical question: determine methods that converge to the functional value l_{n}â†'l(u) much more rapidly than u_{n}â†'u